It is hard to believe that a modern processor (CPU) is the most complex product in the world. What is so complex in this piece of metal?



In this article I will try to explain how a modern processer (CPU) is made from the sand. 

Processor Manufacturing

It takes about $ 5 billion dollars to build a processor manufacturing factory. This factory approximately has 4 years to return the invested funds in its technology, before it will start making the profit. If we make some simple calculations that comes to 100 microchips per hour that the factory should manufacture in order to return the invested funds. 

The process of processor manufacturing looks like this: the special equipment is used to grow a mono-crystal of cylindrical shape from the molten silicon. Next, this resulting ingot is cooled and cut into the wafers, which surfaces are carefully leveled and polished to a mirror shine. In the bio-clean rooms of semiconductor factories are created the micro-circuitries on the silicon wafers using photolithography and etching. Then lab personnel make the random testing of processors under a microscope after re-cleaning of the wafers, and if everything is okay, the finished wafers are sliced into individual processors, which later are put in the casing. 

Some Chemistry Lessons

Let us take a closer look at the whole process of manufacturing. Sand is made up of 25 percent silicon that is the second most abundant chemical element that’s in the Earth’s crust after oxygen. Sand, especially quartz, has high percentages of silicon in the form of silicon dioxide (SiO2) and is the base ingredient for semiconductor manufacturing.



Originally, it is taken in the form of SiO2 sand, which is in arc furnaces (at the temperature about 1800 ° C) reduced by coke:

SiO2 + 2C = Si + 2CO

This silicon is called a "technical" and has a purity of 98-99.9%. For the manufacturing of processors required pure raw material that is called "electronic grade silicon", which should not have more than one alien atom for every one billion silicon atoms. In order to clean it to such level, silicon literally gets "reborn." The silicon tetrachloride (SiCl4) is gotten by chlorination of technical silicon, which subsequently is converted into trichlorosilane (SiHCl3):

3SiCl4 + 2H2 + Si ↔ 4SiHCl3

These recycling reactions that are generated by side silicon-contained materials reduce the cost and eliminate the environmental problems:

2SiHCl3 ↔ SiH2Cl2 + SiCl4
2SiH2Cl2 ↔ SiH3Cl + SiHCl3
2SiH3Cl ↔ SiH4 + SiH2Cl2
SiH4 ↔ Si + 2H2

The silicon enters the melting phase after the purification process. In this picture you can see how one big crystal is grown from the purified silicon melt. The resulting mono-crystal is called an ingot. A mono-crystal ingot is produced from electronic grade silicon. One ingot weighs approximately 100 kilograms (or 220 pounds) and has a silicon purity of 99.9999 percent.



The ingot is then moved onto the slicing phase where individual silicon discs, called wafers, are sliced thin. Some ingots can stand higher than five feet. Several different diameters of ingots exist depending on the required wafer size. Today, CPUs are commonly made on 300 mm wafers. Once cut, the wafers are polished until they have flawless, mirror-smooth surfaces. Intel doesn’t produce its own ingots and wafers, and instead purchases manufacturing-ready wafers from third-party companies. Intel’s advanced 45 nm High-K/Metal Gate process uses wafers with a diameter of 300 mm (or 12-inches). When Intel first began making chips, it printed circuits on 50 mm (2-inches) wafers. These days, Intel uses 300 mm wafers, resulting in decreased costs per chip.



Manufacturing of chips contains more than three hundred operations in which more than 20 layers form a complex three-dimensional structure. So, here we will talk very briefly only about the most important stages.

Photolithography

The problem is solved by using photolithography technology. It is the process of transferring geometric shapes on a mask to the surface of a silicon wafer. This process involves many steps such as:

Photo Resist Application

The blue liquid that is depicted below is a photo resist finish similar to those used in a film for photography. The wafer spins during this step to allow an evenly-distributed coating that’s smooth and also very thin.



UV Light Exposure
At this stage, the photo-resistant finish is exposed to ultra violet (UV) light. The chemical reaction triggered by the UV light is similar to what happens to film material in a camera the moment you press the shutter button.
Areas of the resist on the wafer that have been exposed to UV light will become soluble. The exposure is done using masks that act like stencils. When used with UV light, masks create the various circuit patterns. The building of a CPU essentially repeats this process over and over until multiple layers are stacked on top of each other.
A lens (middle) reduces the mask’s image to a small focal point. The resulting “print” on the wafer is typically four times smaller, linearly, than the mask’s pattern.



More Exposing

In the picture we have a representation of what a single transistor would appear like if we could see it with the naked eye. A transistor acts as a switch, controlling the flow of electrical current in a computer chip. Intel researchers have developed transistors so small that they claim roughly 30 million of them could fit on the head of a pin.

Photo Resist Washing

After being exposed to UV light, the exposed blue photo resist areas are completely dissolved by a solvent. This reveals a pattern of photo resist made by the mask. The beginnings of transistors, interconnects, and other electrical contacts begin to grow from this point.



Etching

The photo resist layer protects wafer material that should not be etched away. Areas that were exposed will be etched away with chemicals.



Photo Resist Removal

After the etching, the photo resist is removed and the desired shape becomes visible.



Re-applying More Photo Resist

More photo resist (blue) is applied and then re-exposed to UV light. Exposed photo resist is then washed off again before the next step, which is called ion doping. This is the step where ion particles are exposed to the wafer, allowing the silicon to change its chemical properties in a way that allows the CPU to control the flow of electricity.



Ion Doping

Through a process called ion implantation (one form of a process called doping) the exposed areas of the silicon wafer are bombarded with ions. Ions are implanted in the silicon wafer to alter the way silicon in these areas conduct electricity. Ions are propelled onto the surface of the wafer at very high velocities. An electrical field accelerates the ions to a speed of over 185,000 mph.



More Photo Resist Removal

After the ion implantation, the photo resist will be removed and the material that should have been doped (green) now has alien atoms implanted.



A Transistor

This transistor is close to being finished. Three holes have been etched into the insulation layer (magenta color) above the transistor. These three holes will be filled with copper, which will make up the connections to other transistors.



Electroplating the Wafer

The wafers are put into a copper sulphate solution at this stage. Copper ions are deposited onto the transistor through a process called electroplating. The copper ions travel from the positive terminal (anode) to the negative terminal (cathode) which is represented by the wafer.



Ion Setting

The copper ions settle as a thin layer on the wafer surface.



Polishing Excess Material

The excess material is polished off leaving a very thin layer of copper.



Layering

Multiple metal layers are created to interconnects (think wires) in between the various transistors. How these connections have to be “wired” is determined by the architecture and design teams that develop the functionality of the respective processor (for example, Intel’s Core i7 processor). While computer chips look extremely flat, they may actually have over 20 layers to form complex circuitry. If you look at a magnified view of a chip, you will see an intricate network of circuit lines and transistors that look like a futuristic, multi-layered highway system.



Testing

Once all of the metal layers are built up, and the circuits (transistors) are all created, it’s time for testing. A device with lots of prongs sits down on top of the chip, attaching microscopic leads to the chip’s surface. Each lead completes an electrical connection within the chip, simulating how it would operate in final form once packaged into end-consumer products.
A series of test signals are sent to the chip with whatever the results are being read. This level of testing includes not only traditional computational abilities, but also internal diagnostics along with voltage readings, cascade sequences (does data flow through as it should), etc. And however the chip responds as a result of this testing, is what’s stored in a database assigned specifically for that die.
This process is repeated for every die on the entire wafer’s surface while all dies are still on the surface.



Wafer Slicing

After tests determine that the wafer has a good yield of functioning processor units, the wafer is cut into pieces (called dies).

Packaging

At this point, all working dies get put into a physical package. It’s important to note that while they’ve had preliminary tested and were found to operate correctly, this doesn’t mean they’re good CPUs.
The physical packaging process involves placing the silicon die onto a green substrate material, to which tiny gold leads are connected to the chip’s pins or ball grid array, which show through the bottom side of the package. On the top of that, a heat spreader is introduced. This appears as the metal package on top of a chip. When finished, the CPU looks like a traditional package end-consumers buy.



A Finished CPU

A microprocessor is the most complex manufactured product on earth. In fact, it takes hundreds of steps and only the most important ones have been visualized in this picture story.




CPU Binning

Based on the test result of class testing processors with the same capabilities are put into the same transporting trays. This process is called “binning,” a process with which many Tom’s Hardware readers will be familiar. Binning determines the maximum operating frequency of a processor, and batches are divided and sold according to stable specifications.



The best chips are generally binned as higher-end parts, being sold as not only the fastest parts with their full caches enabled, but also the low-voltage and ultra low-voltage models.